The present invention is generally directed to a cooling module which is especially useful for either air or water cooling of electronic circuit components, particularly those used in computer systems. More particularly, the present invention is directed to a flexible cooling system for a range of computer systems extending from workstation devices to massively parallel processing system complexes. Even more particularly, the apparatus of the present invention provides increased thermal flexibility and cooling upgradeability to computer installations at lower cost, with greater reliability and without the necessity of system down time.
It has been the trend in customer workstation installations to link a number of computer systems together. Typically, these can take the form of a Local Area Network (LAN) to facilitate cooperative work between the users and to ensure the maximum utilization of each computer system. Additionally, in some installations groups of workstations are being linked together to form loosely coupled parallel processing machines. In the limit, large numbers of workstation processors may be packaged together to form massively parallel supercomputer systems. It is therefore seen that one of the trends in the computer industry is the combination of sub-assemblies of electronic circuit boards into larger systems. Accordingly, it is seen that it is very desirable that these sub-assemblies be configured to facilitate their use in workstations, workstation clusters and also in massively parallel machines. However, the cooling requirements for these various systems are very different. Nonetheless, it would still be desirable to have a cooling methodology which extends across this whole range of computational devices, which is relatively inexpensive to implement and yet which provides flexible and field upgradeable systems.
One of the possibilities for cooling advanced processor circuit modules is the employment of arrays of parallel plates (fins) which may be employed as heat sinks. Heat generated in an electronic component is conducted into the plates and dissipated through the passage of a forced flow of ambient air within high aspect ratio flow channels between the fins. In future machines, as power levels per module increase, the combined heat dissipated by many machines in a confined work space, whether they are independent or part of a large LAN, could exceed the capacity of the room air conditioning system in which the systems are placed. In this case, it would be desirable if the installation could employ an alternative of a liquid cooling mechanism for these workstations using chilled water from a central Coolant Distribution Module (CDM). The option of air or water cooling for the sub-assemblies is also desirable for cooling massively parallel machines. In such cases, the cooling demands are often present from the point of initial system installation. However, as the number of processors is expanded, even in massively parallel machines, it is seen that it is desirable to have flexibility in terms of cooling capacity. While the total heat load generated from smaller machines may well be dissipated into the ambient atmosphere, and eventually controlled by means of a building air conditioning system, the heat load generated by a large machine, containing hundreds or even thousands of processor sub-assemblies packaged into a small volume, would almost certainly be too large (hundreds of kilowatts) to dissipate directly to the room air environment.
It is also noted that from a thermal perspective, thin plates used as heat sinks cannot be simultaneously optimized for both air and water (or other liquid or refrigerant) cooling. This limitation imposes additional constraints upon the design of any computer cooling system which is meant to span a range of system performance levels (workstations at the low end to massively parallel processors at the high end).